A neural network model applied to the detection of digital signals

This work presents an artificial neural network (ANN) approach for the signal decision problems associated with digital communication system receivers which use modulation schemes whose signal elements belong to finite bidimensional constellations. The decision system proposed, named a neural decoder (ND), is a multilayer perceptron neural network trained with a backpropagation algorithm and models a maximum-likelihood receiver. The ND training process and simulation results of their performance, regarding a conventional receiver, are presented for some of the modulation systems studied.